In [1]:

    

%matplotlib inline


    

In [2]:

    

import matplotlib.pyplot as plt


    

In [3]:

    

pathToImage = '/Users/nick/Downloads/neurofinder.02.00/images_summary/mean.tif'
pathToSources = '/Users/nick/Downloads/neurofinder.02.00/regions/regions3.json'
pathToSourcesGT = '/Users/nick/Downloads/neurofinder.02.00/regions/regionsRand.json'


    

In [4]:

    

import extraction as ex


    

In [5]:

    

model = ex.load(pathToSources)


    

In [6]:

    

from skimage.io import imsave, imread


    

In [7]:

    

im = imread(pathToImage)


    

In [8]:

    

from showit import image


    

In [54]:

    

from numpy import tile, maximum


    

In [218]:

    

base = tile((im.astype(float)/600).clip(0,1),(3,1,1)).transpose(1,2,0)
masks = model.regions.mask((512,512), background='black', fill=[0.8, 0, 0], stroke='white')
blend = maximum(base, masks)


    

In [219]:

    

fig = plt.figure(figsize=[10,10])
ax = plt.axes()
image(blend, ax=ax)
plt.xlim([0, blend.shape[1]]);
plt.ylim([blend.shape[0], 0]);
#for s in range(randR.regions.count):
#    plt.annotate(s=str(s), xy=[model.regions.center[s][1],model.regions.center[s][0]], color='w', size = 10);


    

In [84]:

    

imsave(pathToImage[:-4] + '-sources.tif', (255*blend).astype('uint8'), plugin='tifffile', photometric='rgb')


    

In [117]:

    

import neurofinder as nf


    

In [223]:

    

threshold = 5


    

In [124]:

    

matches = nf.match(model.regions, modelRand.regions, threshold)


    

In [220]:

    

recall, precision = nf.centers(model.regions, modelRand.regions, threshold)


    

In [221]:

    

inclusion, exclusion = nf.shapes(model.regions, modelRand.regions, threshold)


    

In [224]:

    

d = {'recall':recall, 'precision':precision, 'inclusion':inclusion, 'exclusion':exclusion, 'threshold':threshold}


    

In [227]:

    

import json
from os.path import join


    

In [228]:

    

output = '/Users/nick/Downloads/neurofinder.02.00/regions'
with open(join(output, 'results.json'), 'w') as f:
    f.write(json.dumps(d, indent=2))


    

In [155]:

    

from regional import many


    

In [170]:

    

from numpy import where, isnan, nan, full


    

In [181]:

    

matchesRR = full(modelRand.regions.count,nan)
for ii in where(~isnan(matches))[0]:
    matchesRR[matches[ii]] = ii


    

In [187]:

    

matchedA = ex.model.ExtractionModel([model.regions[i] for i in where(~isnan(matches))[0]])
matchedB = ex.model.ExtractionModel([model.regions[i] for i in where(isnan(matches))[0]])
matchedC = ex.model.ExtractionModel([modelRand.regions[i] for i in where(~isnan(matchesRR))[0]])
matchedD = ex.model.ExtractionModel([modelRand.regions[i] for i in where(isnan(matchesRR))[0]])


    

In [216]:

    

base = tile((im.astype(float)/600).clip(0,1),(3,1,1)).transpose(1,2,0)
masksA = matchedA.regions.mask((512,512), background='black', fill='green', stroke='black')
masksB = matchedB.regions.mask((512,512), background='black', fill='red', stroke='black')
masksC = matchedC.regions.mask((512,512), background='black', fill='green', stroke='white')
masksD = matchedD.regions.mask((512,512), background='black', fill=[.7, 0, 0], stroke='white')
X = maximum(maximum(maximum(masksA, masksB), masksC), masksD)
blend = maximum(base, X)


    

In [217]:

    

fig = plt.figure(figsize=[10,10])
ax = plt.axes()
image(blend, ax=ax)
plt.xlim([0, blend.shape[1]]);
plt.ylim([blend.shape[0], 0]);
#for s in range(randR.regions.count):
#    plt.annotate(s=str(s), xy=[model.regions.center[s][1],model.regions.center[s][0]], color='w', size = 10);


    

In [86]:

    

from regional import one, many
from showit import image
from numpy import zeros, random, asarray, round, where, ones
from scipy.ndimage.morphology import binary_closing, binary_opening, binary_fill_holes, binary_dilation


    

In [108]:

    

dims = [512,512]
margin = 20
n = 100


    

In [109]:

    

def topoly(c):
    tmp = zeros(dims)
    coords = asarray([c[0] + random.randn(32) * 3, c[1] + random.randn(32) * 3]).astype('int')
    tmp[coords.tolist()] = 1
    tmp = binary_dilation(tmp, ones((3, 3)))
    tmp = binary_closing(tmp, ones((7, 7)))
    return asarray(where(tmp)).T

xcenters = (dims[0] - margin) * random.random_sample(n) + margin/2
ycenters = (dims[1] - margin) * random.random_sample(n) + margin/2
centers = zip(xcenters, ycenters)


    

In [110]:

    

regions = many([one(topoly(c)) for c in centers])


    

In [111]:

    

modelRand = ex.model.ExtractionModel(regions)